A Rust lang team member reflects on how the Rust language design process suffers from unbounded queues — too many open RFCs and features in flight with no capacity constraints. The post argues that accepting RFCs without ensuring resources are available leads to features languishing for years (e.g., RFC 66 from 2014), confuses users about priorities, and burns out contributors. The author advocates for introducing a hard cap on concurrent active work items, explicit 'tabling' of deferred work, and measuring capacity in terms of reviewer bandwidth rather than contributor count. The ideas are compared to Kanban-style workflow management.
Table of contents
The feature pipelineToday we have unbounded queuesHow can an RFC sit around for 5 years?Unbounded queues lead to confusion for usersUnbounded queues make it harder for teamsLeaving room for serendipity?Reviewing as the limited resourceLeaving room for new ideas?Tabling as an explicit actionPeople need help to set limitsConclusionFootnotes1 Impression